The EMBO Journal vol. 1 1 no.7 pp.2407 - 2414, 1992

Influenza virus hemagglutinin with multibasic cleavagesite is activated by furin, a subtilisin-like endoprotease

Andrea Stieneke-Grober, Martin Vey, HerbertAngliker', Elliott Shawl, Gary Thomas2,Christopher Roberts, Hans-Dieter Kienk andWolfgang Garten

Institut fur Virologie, Philipps-Universitat Marburg, Robert-Koch-Strasse 17, D-3550 Marburg, Germany, 'Friedrich Miescher-Institute,PO Box 2543, CH-4002, Basel, Switzerland and 2Vollum Institute,Oregon Health Sciences University, Portland, OR 97201-3098, USA

Communicated by H.-D.Klenk

Many viruses have membrane glycoproteins that areactivated at cleavage sites containing multiple arginineand lysine residues by cellular proteases so far notidentified. The proteases responsible for cleavage of thehemagglutinin of fowl plague virus, a prototype of theseglycoproteins, has now been isolated from Madin-Darbybovine kidney cells. The enzyme has a mol. wt of 85 000,a pH optimum ranging from 6.5 to 7.5, is calciumdependent and recognizes the consensus sequence R-X-K/R-R at the cleavage site of the hemagglutinin. Usinga specific antiserum it has been identified as furin, asubtilisin-like eukaryotic protease. The fowl plague virushemagglutinin was also cleaved after coexpression withhuman furin from cDNA by vaccinia virus vectors.Peptidyl chloroalkylketones containing the R-X-K/R-Rmotif specifically bind to the catalytic site of furin andare therefore potent inhibitors of hemagglutinin cleavageand fusion activity.Key words: fowl plaque virus/furin/hemagglutinin/multibasiccleavage site/subtilisin-like endoproteases

IntroductionEndoproteolytic cleavage at arginine residues is a commonpost-translational modification of membrane and secretoryproteins on the exocytotic transport route. Such proteinsinclude precursors of peptide hormones, neuropeptides,growth factors, coagulation factors, serum albumin, cellsurface receptors and adhesion molecules (for review seeBarr, 1991).The available evidence indicates that the cellular proteases

involved in the processing of these proteins are alsoresponsible for the cleavage of many viral membraneproteins. Detailed information on the biological significanceof proteolytic cleavage in a virus system has been obtainedfrom studies on the hemagglutinin (HA) of influenza virus(for review see Klenk and Rott, 1988). Cleavage of the HA,which is essential for the ability of the virus to enter cells,proved to be an important determinant for the spread ofinfection through the organism and for virus pathogenicity.

Oxford University Press

The mammalian and the apathogenic avian influenza virusstrains have HAs that are cleaved only in a restricted numberof cell types. These viruses therefore cause local infection.On the other hand, the pathogenic avian strains have HAsactivated in a broad range of different host cells and, thus,cause systemic infection. There are several structuralproperties of the HA that determine the differentialcleavability, but the key factor is the amino acid sequenceat the cleavage site (Kawaoka and Webster, 1988; Ohuchiet al., 1989, 1991; Khatchikian et al., 1989; Li et al., 1990).Mammalian and apathogenic strains have a single arginineat this site, and plasmin (Lazarowitz and Choppin, 1975),a factor X-like protease from allantoic fluid of chicken eggs(Gotoh et al., 1990; Ogasawara et al., 1992) and bacterialproteases (Tashiro et al., 1987) have been identified asenzymes activating this type of HA. The HAs of thepathogenic strains, on the other hand, have long been knownto contain multiple lysine and arginine residues at theircleavage sites (Bosch et al., 1981; Kawaoka and Webster,1988), and recently it has been shown that the consensussequence R-X-K/R-R is critical for cleavage activation (Veyet al., 1992).The ubiquitous proteases responsible for the activation of

the HA of pathogenic strains were not well understood.However, it was known that they are calcium dependent andhave a neutral pH optimum (Klenk et al., 1984), and thatthey can be inhibited by specific peptidyl chloroalkylketones(Garten et al., 1989). Such enzymes appear to be highlyconserved, since the HA of fowl plague virus (FPV), theprototype of these pathogenic avian strains is activated notonly in virtually all mammalian and avian cells analyzed butalso in invertebrate cells (Kuroda et al., 1986). It wastherefore interesting to find that a protease resembling theseenzymes in its catalytic and other biochemical propertiesexists in the yeast Saccharomyces cerevisiae (Fuller et al.,1988). This Kex2 protease cleaves its natural substrates, pro-a-factor and pro-killer toxin, at R-R and K-R sites, but itcan also process mammalian hormone and neuropeptideprecursors. Furthermore, several Kex2 analogues,furin/PACE (Fuller et al., 1989; Hatsuzawa et al., 1990;Wise et al., 1990), PC2 (Seidah et al., 1990; Smeekens andSteiner, 1990) and PC1/PC3 (Seidah et al., 1991; Smeekenset al., 1991) have recently been discovered in human andmouse tissues. These enzymes together constitute a newfamily of eukaryotic subtilisin-like endoproteases (Barr,1991). In the present study we show that human furinexpressed from cDNA activates the FPV HA by proteolyticcleavage. Furthermore, the activating enzyme present inMadin-Darby bovine kidney cells, a cell line allowing FPVreplication in vivo, has been identified as a furin-likeprotease. These findings indicate that the proteasesresponsible for the activation of viral glycoproteins at

multibasic cleavage sites are subtilisin-like enzymes.

2407

A.Stieneke-Grober et al.

ResultsFPV HA is cleaved by human furin expressed underthe direction of a vaccinia virus recombinantFor the reasons outlined above, furin appeared to be a likelycandidate for an activating enzyme of the FPV HA. To provethis hypothesis, HA cDNA and human furin cDNA isolatedfrom a hepatoma (HepG2) cDNA library (Bresnahan et al.,1990; Wise et al., 1990) were expressed in CV-l cells aloneor in combination, using vaccinia virus recombinants asvectors (Figure 1). Lanes 1 and 2 demonstrate that, whenexpressed alone or together with the HA, furin can bedetected as a 90-96 kDa band after immunoprecipitationas has been described before (Wise et al., 1990). When HAexpressed alone was analyzed by pulse-chase labeling at4-6 h after infection, it was detected mainly in the formof subunits HA, and HA2, since the precursor was rapidlycleaved by the proteolytic activity endogenous to CV-1 cells.However, in agreement with a previous report on murine3-nerve growth factor (Bresnahan et al., 1990), there was

an accumulation of the precursor HA late after infection.Thus, when pulse -chase labeled at 20 h post-infection, HAexpressed alone persisted in the uncleaved form (Figure 1,lane 3). It was therefore possible to analyze the effect ofover-expressed furin on HA processing. The results of anexperiment in which HA was labeled in the same way aftercoexpression with furin are shown in lane 4. The observa-tion that the HA is now predominantly present in the leavedform demonstrates that furin of human origin is able toprocess this glycoprotein.

Identification of a furin-like protein as the activatingenzyme of MDBK cellsIt was therefore of interest to find out if the endogenousprotease responsible for HA activation in MDBK cellsinfected with FPV was related to furin, with which it sharedsuch characteristics as calcium dependence and neutral pHoptimum as pointed out above. Although it was known fromour previous study that lysates of MDBK cells contained aHA activating enzyme (Klenk et al., 1984), we were unableto detect furin in these lysates by the Western blot techniqueemploying an immune serum that was generated against the

Fig. 1. Coexpression of human furin and HA b recombinant vacciniaviruses. CV-1 cells were labeled for I h with [ 5S]methionine(100 /Ci/mI) at 20 h p.i. with recombinant vaccinia viruses VVhFurand VV-HAwt, expressing the fir gene (Bresnahan et al., 1990) andthe HA gene of influenza virus A/FPV/Rostock/34 (H7N1)respectively. After I h chase with unlabeled methionine the cells were

lysed, the vectorially expressed proteins immunoprecipitated with therespective antisera and analyzed by SDS -PAGE and fluorography.Lane 1, furin immunoprecipitated from cells infected with VVhFur andVV-HAwt; lane 2, furin immunoprecipitated from cells infected withVVhFur; lane 3, FPV HA immunoprecipitated from cells infected withVV-HAwt; lane 4, FPV HA immunoprecipitated from cells infectedwith VVhFur and VV-HAwt.

2408

catalytic domain of human furin expressed as a fusion proteinin Escherichia coli (Wise et al., 1990). Since the failure todetect furin by this approach might have reflected too lowconcentrations in MDBK cells, an attempt was made topurify furin and to monitor it in the enriched enzymepreparation (Table I). Three independent assays were usedfor enzyme identification throughout the purificationprocedure. These were cleavage analysis of the FPV HAusing virions containing the precursor HA as substrate (Kienket al., 1984), cleavage analysis using the fluorogenic peptideRQRR-AMC which mimics as substrate the cleavage siteand immune analysis employing the described immuneserum.The first step in the isolation procedure involved

fractionation of cytoplasmic extracts of MDBK cells ondiscontinuous sucrose density gradients. Cell fractionscontaining plasma membrane, lysosomes, Golgi apparatusand endoplasmic reticulum have been characterized byspecific marker enzymes. HA cleavage activity cosedimentedwith the Golgi fraction which presumably also included thetrans-Golgi network, suggesting that the activating enzymeis located in one of these organelles (data not shown). Sinceit was known that the activating protease is membrane bound(Klenk et al., 1984), n-octylglucoside was used for thesolubilization of the Golgi fraction and employed in allfurther purification steps. Another essential factor in theisolation procedure was the use of the protease inhibitorsleupeptin, E-64, PMSF and pepstatin. As shown below

Table I. Purification of HA cleaving protease from MDBK cells

Purification step Activity Total protein Sp. act.(units) (mg) (units/mg)

1. Microsomes 13 401 102.3 1312. Sucrose gradient 12 015 53.4 2253. DEAE-Sepharose CL6B 7830 8.7 9004. Mono Q FPLC 1691 0.29 5.8335. Superose 12 FPLC 600 0.057 10.526

1 unit: 1 ng HA cleaved per minute at 37°C.

Table II. Effect of inhibitors on HA cleaving protease of MDBK cells

Inhibitors Concentration Inhibition ofHA cleavage (%)

DFP 1 mM 0PMSF 1 mM 0TLCK 1 mM 0Leupeptin 1 mM 0Aprotinin 1 mM 0lodoacetamide 1 mM 0N-Ethylmaleimide 1 mM 0Cadmium 1 mM 0Cobalt 1 mM 0E64 1 mM 0Hydroxymercuribenzoic acid 1 mM 80EDTA 50 mM 100EGTA 50 mM 100Phenanthroline 1 mM 0Pepstatin A 50 ug/ml 0

Lysates of microsomes from MDBK cells were pre-incubated with theinhibitors in the presence of 10 mM CaCI2. Aliquots of the pre-treatedlysates were then incubated at 37°C for 30 min with [35S]methionine-labeled FPV containing uncleaved HA. The incubation mixtures weredissolved in sample buffer and analyzed by SDS-PAGE. Cleavage ofHA was densitometrically estimated from fluorographies.

Hemagglutinin activation by furin

(Table II), these inhibitors do not interfere with the HAactivating enzyme, when analyzed in fresh cell lysates. Theyinactivate, however, other proteases that may degrade theHA activating protease or the virus used as substrate in thecourse of the isolation procedure. In the second step of thisprocedure, the solubilized Golgi proteins were separated byultracentrifugation in a continuous sucrose density gradient,followed by anion exchange chromatography in step 3.In step 4, the material was subjected to Mono Q FPLC.Figure 2 demonstrates that the fractions containing the HAactivating protease also cleave the fluorogenic tetrapeptideRQRR-AMC, whereas another proteolytic activity specificfor RR-AMC has been separated. The fluorogenic peptideRVRR-AMC had the same specificity as RQRR-AMC, butwas - 5-fold more effective (data not shown). In the finalstep involving gel filtration by Superose 12 FPLC furtherpurification of the HA activating protease was obtained(Figure 3). These results therefore indicate that MDBK cellscontain, in addition to the protease activating HA, otherproteases specific for multiple basic residues that have beenseparated during the purification procedure. As has beenpointed out above, the protease fractions obtained throughoutthe isolation procedure have been monitored for theirimmunoreactivity with an antiserum specific for human furinby Western blot analysis. Whereas furin-related proteinscould not be detected by this approach in MDBK cell lysatesor at the initial purification steps, a single protein band (mol.wt 85 000) was observed after Mono Q and Superose 12

FPLC that was copurified with HA cleaving activity (Figures2 and 3). It has to be pointed out that this protein has asmaller mol. wt (85 000) than human furin expressed inCV-1 cells (96 000) (Figure 1). These results indicate thatMDBK cells contain a furin-related protein and they stronglysuggest that this protein activates HA.Our previous findings had already indicated that correct

cleavage of the FPV HA by crude MDBK lysates occurredonly at neutral pH (Klenk et al., 1984). We have thereforedetermined the pH optimum of the enzyme preparationobtained after Superose 12 chromatography using RQRR-AMC as substrate. As shown in Figure 4, highest activitywas observed between pH 6.5 and 7.5. This finding supportsthe concept that the FPV HA is activated in MDBK cellsat pH values close to neutrality.

Endopeptidases are divided into four classes, serine- (EC3.421), cysteine- (EC 3.422), aspartate- (EC 3.423) andmetalloproteinases (EC 3.424), which can be discriminatedby their sensitivity to specific inhibitors. We have analyzedthe effect of these inhibitors on cleavage of the HA by theMDBK cell protease (Table II). Cleavage was inhibited byEDTA and EGTA reflecting the calcium dependence of theenzyme (Klenk et al., 1984). However, the third inhibitorof metalloproteinases, phenanthroline, was ineffective. Ofthe inhibitors specific for cysteine proteinases, iodo-

s - 'oR&_ #S W .,*-:4ft J ..* .. iv

Fig. 2. Ion exchange chromatography of the HA activating protease ofMDBK cells. Fractions containing protease activity obtained afterDEAE-Sepharose CL6B chromatography were loaded on a Mono Qcolumn. Proteins were separated by FPLC using a 0-300 mM NaCIgradient. Upper panel, protease activities of each fraction were testedby incubation with the fluorogenic peptidyl-7-amido-3-methylcoumarinsindicated. Middle panel, protease activity was monitored by cleavageof HA after incubation of fractions from Mono Q with[35S]methionine-labeled A23187-FPV. Viral proteins were analyzed bySDS-PAGE and fluorography. Lower panel, aliquots of Mono Qfractions were subjected to SDS-PAGE, transferred to a nitrocellulosemembrane and analyzed by immunoreaction using antiserum againstfurin.

Fig. 3. Gel filtration of the HA activating protease MDBK cells.Protease fractions collected from the Mono Q column were separatedby FPLC on a Superose 12 column. Eluted fractions (I ml) of thechromatography containing HA cleaving activities are shown. Upperpanel, protein profile of the eluate (solid line) and mol. wt standards(in kDa dextran blue 3000, ferritin 440, catalase 232, aldolase 158,albumin 67, ovalbumin 43. chymotrypsinogen 25, RNase 13.7; dashedline) were measured by optical density at 260 nm. Middle panel,protease activity was monitored by analyzing the cleavage of HA using[35S]methionine-labeled A23187-FPV. Lower panel, aliquots of FPLCfractions were subjected to SDS-PAGE and immunoblotted withantiserum against furin.

2409

6...

qw

A.Stieneke-Grober et at.

0UC

0

0

6,

20.0

15.0-

10.0-

5.0-

5.0 5.5 6.0 6.5 7.0 7.5 8.0 pH

Fig. 4. pH optimum of the HA activating protease of MDBK cells.The activity of the protease obtained after FPLC on Superose 12 wasdetermined in 100 mM sodium acetate buffer, ranging between pH 5.0and 6.5, and 100 mM HEPES- and 100 mM Tris buffer rangingbetween pH 6.5 and 8.0. Boc-Arg-Gln-Arg-Arg-AMC was used assubstrate. The relative fluorescence indicates the proteolytic activity.

acetamide, N-ethylmaleimide, cobalt chloride, cadmiumacetate, E64 and hydroxymercuribenzoate, only the last onewas effective. Neither pepstatin, an inhibitor for aspartateproteinases, nor the serine protease inhibitors DFP, PMSF,TLCK, TPCK, leupeptin and aprotinin interfered withcleavage. The ineffectiveness of PMSF was not expectedsince this compound has been reported to inhibit human furin(Bresnahan et al., 1990). These results indicate that theMDBK cell protease, although a furin-like enzyme and thusbelonging to the serine proteinases, does not have theinhibitor specificity typical for this class of endopeptidases.

Inhibition of cleavage activation by sequence-specificpeptidyl chloromethylketonesWe have previously shown that, unlike peptidylchloromethylketones containing a single arginine,compounds with paired basic amino acids inhibit cleavageactivation of the FPV HA and, thus prevent multiple cyclesof virus replication (Garten et al., 1989). Since it becameclear that the consensus sequence has an additional argininefour amino acids upstream of the cleavage site (Vey et al.,1992), peptidyl chloromethylketones with an arginine in thisposition have now also been synthesized and tested for theiractivity to inhibit cleavage by the MDBK cell protease. Mostof these inhibitors were acylated to increase cellular uptake(Garten et al., 1989). In the experiment shown in Figure 5,we analyzed the effect of such inhibitors using A23187virions with uncleaved HA that had been incubated in vitrowith the protease preparation obtained after Superose 12chromatography. As shown before, the dibasic compounddecFAKR-CMK inhibits at a 1 mM concentration. The newcompound decREIR-CMK which has a single basic residuenext to the cleavage site but an additional arginine at position-4 resembles in its inhibitory potential the other dibasiccompound. However, when a pair of basic residues atpositions -1 and -2 is combined with an arginine at position-4, as is the case with decREKR-CMK, there was anincrease in sensitivity by a factor of - 1000.We have also analyzed the effect of these inhibitors on

in vivo cleavage of HA in FPV-infected MDBK cells (Figure6). Again decREKR-CMK is the most efficient inhibitor.The data show also that distinct cleavage inhibition, asoccurring at 100 ,uM decREKR-CMK, is not paralleled bya comparable reduction of hemagglutination titre in themedium. Thus, - at appropriate concentrations, peptidylchloromethylketones inhibit cleavage without interfering withparticle formation. Two analogues ofdecFAKR-CMK were

Fig. 5. Inhibition of the HA activating protease of MDBK cell byspecific decanoylated peptidyl chloromethylketones. Purified HAactivating protease was incubated with the inhibitors for 30 min at37°C and analyzed for residual HA cleaving activity usingA23187-FPV as substrate. Fluorographies of viral proteins (upperpanel) and the densitometrical evaluation of HA cleavage as shown(lower panel).

also analyzed: decFAKR-S+, a proteinase inhibitor (Shaw,1988), in which the chloromethylketone group had beenreplaced by a sulfonium group, and palFAKR-CMK, inwhich the fatty acid had been changed from decanoic topalmitic acid. As shown in Figure 6, neither modificationimproved the inhibitory effect ofdecFAKR-CMK. It shouldbe pointed out that - 100-fold higher concentrations ofdecREKR-CMK had to be used here than in experimentswhere the inhibitory effects on in vitro cleavage (Figure 5)and on fusion activity (see below, Figure 7) were analyzed.This may be explained by the instability ofdecREKR-CMKwhich requires relatively high concentrations in long-timeexperiments like this one.

It was then of interest to find out if the inhibitory effectof the peptidyl chloromethylketones also interfered with thebiological activities of the HA. To this end, polykaryocytosisin BHK 21-F cells induced by HA at low pH was analyzedas depending on inhibitor concentration. Figure 7demonstrates that complete inhibition of cell fusion was againobtained at micromolar concentrations of decREKR-CMK,whereas decFAKR-CMK and decRAIR-CMK had to bepresent at 100- to 1000-fold higher concentrations to producethe same effect. It is therefore clear that the peptidylchloromethylketones with the correct amino acid sequence,while not interfering with hemagglutination and thusreceptor-binding (Figure 6), specifically block the cleavage-dependent fusion activity of the HA.Taken together these data indicate that decREKR-CMK

is the most potent one of the inhibitors of HA cleavageanalyzed. The amino acid sequence of this inhibitor is, inessence, identical to that of the fluorogenic compoundRQRR-AMC which, as shown in Figure 2, is a specific

2410

.16-460460-Ir4.4-1.0 1.1-op-, -1-11- 1-1.11-1001,

4 .,-a:: .., -;

- .-Aam-,Nm-- -jush, ampi" .Iqjrl -9wOr. 4s0

Al" AMb.a6l;.%-lob bas

Hemagglutinin activation by furin

4.. R

I0

2

u.

rA

NPHA.

HA2M

Fig. 6. Effect of various peptidyl chloroalkylketones on HA cleavage in cells infected with FPV. Inhibition of HA cleavage by the acylated peptidylchloroalkylketones and the sulfonium compound was analyzed at 10, 100 and 1000 FM concentrations respectively, in chick embryo cells infectedwith FPV/Rostock/34 (H7N1) (109 p.f.u./ml). The cleavage patterns of the FPV HA were analyzed in infected cells by SDS-PAGE andfluorography. Hemagglutination titers of virus released into the medium are also indicated.

substrate for the furin-like protease from MDBK cells. Theseobservations strongly suggest that decREKR-CMK interfereswith cleavage activation of the FPV HA, because itspecifically inhibits this protease. To strengthen further thisconcept, a radioiodinated peptidyl chloromethylketone of thestructure [1251]YARAKR-CMK was synthesized whichcould be used for affinity labeling studies. Figure 8 showsan experiment in which the fractions obtained from theSuperose 12 column in the final purification step of theMDBK protease were analyzed with this probe. Two majorbands (85 and 69 kDa) were radiolabeled (Figure 8, middlepanel). The 85 kDa protein was identified as the furin-likeprotease by immunoprecipitation (Figure 8, lower panel).The 69 kDa band was present throughout the gradient andhas also been found when affinity label not coupled to proteinwas analyzed. None of the minor bands seen comigrated withthe protease activity. These observations indicate that, in theenzyme preparation obtained from MDBK cells, only thefurin-like protein is able to recognize specifically the cleavagesite of the HA. Thus, clear evidence has been obtained thatthis protein which has been identified now as bovine furinis responsible for HA activation in MDBK cells.

DiscussionWe show here that the HA of FPV, a viral membraneglycoprotein with a multibasic cleavage site, is activated byfurin. This conclusion was reached by two differentapproaches. Firstly, we could demonstrate that the FPV HAwas cleaved, when it was coexpressed with human furinusing vaccinia virus vectors. Expression from cDNA wasalso used to show that furin of human and murine originis able to activate three secretory proteins: von Willebrandfactor (Wise et al., 1990), renin (Hosaka et al., 1991), and13-nerve growth factor (Bresnahan et al., 1990). Secondly,bovine furin was identified as the endogenous proteaseresponsible for cleavage activation of HA in MDBK cells,a bovine kidney cell line allowing in vivo replication of FPV.Since furin is present in MDBK cells in amounts below thethreshold of detection, the enzyme had to be enriched by

EX] dec RAIR-CMK__dec FAKR-CMK- dec REKR-CMK

100

x

U

'a

0

00.el0

.X

(L

1 mM 0.1mM 0.01 mM 0,001 mM

Fig. 7. Effect of peptidyl chloromethylketones on fusion of BHK21-Fcells induced by FPV. Confluent monolayers of BHK21-F cells wereinfected with FPV/Dutch/27 (H7N7) (Dobson variant) and incubatedfor 6 h with the peptidyl chloromethylketones at the concentrationsindicated. The cells were then kept for 2 min at pH 5.5 and analyzedfor polykaryon formation as described in Materials and methods. Thepolykaryocytosis indices (PI) (percentage of total nuclei of a monolayerpresent in polykaryons) are indicated. PI of infected cells withoutinhibitor: 295. PI of uninfected cells: <5.

conventional cell fractionation and protein purificationprocedures. Although purification to homogeneity has notbeen accomplished, the protease was sufficientlyconcentrated to allow unambiguous identification by aspecific antiserum and by affinity labeling with a radioactivepeptidyl chloromethylketone mimicking the cleavage site ofthe HA. We have, thus, not only shown that furin over-expressed from a foreign gene can activate at R-X-K/R-Rcleavage sites, but we have demonstrated, to our knowledgefor the first time, that the indigenous enzyme responsiblefor this type of proteolytic activation in a given cell is furin.As a member of the family of eukaryotic subtilisin-like

endoproteases, bovine furin shows many features typical forthese enzymes. These include, above all, the substratespecificity for multibasic cleavage sites which has beenanalyzed here by using peptidyl coumarins as fluorogenic

2411

io, V,.

.AMWML-:. :.,,...W~

iW.,IF,

-... Ammwffiim..

0'. 40:4.. 40 Is. ID

A.Stieneke-Grober et al.

Fig. 8. Affinity labeling of the HA activating protease of MDBK cellswith radioactive peptidyl chloromethylketone. Aliquots of 0.5 mlfractions eluted from the Sepharose 12 FPLC column were incubatedat 37°C for 30 min with ["5I]YARAKR-CMK at a concentration of0.05 AM and a sp. radioact. of 2000 Ci/mmol, immunoprecipitatedwith anti-furin serum, and subjected to SDS-PAGE. Upper panel,HA protease activity was monitored using Boc-Arg-Val-Arg-Arg-AMCas substrate. Middle panel, column fractions were analyzed bySDS-PAGE followed by autoradiography. Lower panel, fractionswere immunoprecipitated with the anti-furin serum prior toSDS-PAGE.

substrates and by peptidyl chloroalkylketone inhibitors. Theresults indicate that bovine furin specifically recognizes thecleavage sites of the structure R-X-K/R-R, as has beenobserved before with murine furin (Hosaka et al., 1991).These observations are also in agreement with the conceptthat influenza virus HA requires this consensus sequence forhigh cleavability (Vey et al., 1992). Other features sharedby bovine furin and the other subtilisin-like proteinases arecalcium dependence, neutral pH optimum, sensitivity tohydroxymercuribenzoate and insensitivity to many inhibitorsof other serine proteinases. On the other hand, bovine andhuman furin differ in their mol. wt. There are alsodifferences in substrate specificity between the subtilisin-likeproteases. Furin differs from Kex2, the prototype of theseenzymes, by the requirement of arginine at position -4, inaddition to the basic residues at positions -1 and -2 of thecleavage site (Barr, 1991), and studies with yeast extractshave shown that Kex2 is unable to cleave the FPV HA (datanot shown). The differential substrate specificity alsoseparates furin from PC2 and PC1/PC3 which again cleaveonly at dibasic cleavage sites (Hosaka et al., 1991). PC2differs also in its pH optimum at 5.5 (Shennan et al., 1991).

Unlike PC2 and PC1/PC3 which are believed to beresponsible for cleavage of pro-hormones and neuropeptidesin secretory vesicles of the regulated secretory pathway, furinappears to be a component of the constitutive secretorypathway (Barr, 1991). It has long been known that cleavage

of the FPV HA is a late post-translational process, but stilloccurs within the cell (Klenk et al., 1974). More recentlyevidence has been obtained that the trans-Golgi network maybe the cleavage compartment (De Curtis and Simons, 1989).Although the data presented here do not allow an exactallocation of the cleaving enzyme, they are compatible withthe concept that the Golgi complex may be involved.Evidence has also been obtained that the glycoprotein of Roussarcoma virus (Perez and Hunter, 1987), the prM proteinof flaviviruses (Randolph et al., 1990), the F proteins ofmumps (Yamada et al., 1988) and Newcastle disease virus(NDV) (Morrison et al., 1985) are activated in the Golgiapparatus or the trans-Golgi network. Furthermore, acalcium-dependent protease capable of activating the Fprotein of NDV has been isolated from trans-Golgimembranes of liver cells, but its identity has not beenestablished yet (Sakaguchi et al., 1991). The neutral pHoptimum of furin and the requirement of pH 7 for HAcleavage might argue against the Golgi apparatus or thetrans-Golgi network as the cleavage compartment, since theintracistemal milieu of these organelles has been reportedto be acidic (Anderson and Pathak, 1985; Griffith andSimons, 1986). However, firstly it should be pointed outthat the pH optimum of furin is broad, ranging from pH 6.5to 7.5. Furthermore, evidence has been obtained recentlythat at low pH the FPV HA undergoes conformationalalterations interfering with correct cleavage and that, ininfected cells, ion channels formed by the M2 protein ofinfluenza virus raise the pH in the trans-Golgi network(Sugrue et al., 1990). The virus may therefore be able toadjust the intracisternal milieu to optimal conditions for HAprocessing.FPV is the prototype (Klenk and Rott, 1988) of a long

list of other viruses that also contain glycoproteins withR-X-K/R-R cleavage sites. Among these viruses, whichbelong to many different taxonomic families, are importantpathogenic agents, such as measles virus (Richardson et al.,1986), mumps virus (Waxham et al., 1987), caninedistemper virus (Barrett et al., 1987), respiratory syncytialvirus (Collins et al., 1984; Elango et al., 1985), yellow fevervirus (Rice et al., 1985), HIV (Ratner et al., 1985), humancytomegalovirus (Spaete et al., 1990) and varicella-zostervirus (Keller et al., 1986). With many of these virusesevidence has been obtained that glycoprotein cleavage is ofbiological significance, as is the case with influenza virus.It is therefore fair to assume that furin plays an importantrole in the life-cycle of these viruses and in their interplaywith the host. On the other hand, we have presented evidencehere that MDBK cells contain, in addition to furin, otherproteases recognizing multibasic cleavage sites. Althoughthese activities proved to be different from the proteasecleaving the HA, it cannot be ruled out that they activateother viral glycoproteins. Furthermore, there is evidence forvariations of furin between species, and they may also existon the level of tissues and cells. Obviously, such variationsmay be important for organ tropism, host range andpathogenicity of any of these viruses. It also remains to beseen if infections with these viruses can be controlled by thepeptidyl chloromethylketones. Since these inhibitors are alsolikely to interfere with the biosynthesis of many importantcellular proteins, they are usually considered toxic. However,we have already shown previously that dibasicchloromethylketones, when interfering with FPV replication

2412

.... ...... 44-.. t..-.,5A

*0 :Mkm&.

.: II .-.

..-- AW.4w

Hemagglutinin activation by furin

in tissue culture, cause relatively little cell damage (Gartenet al., 1989). Since our new compounds containing the fullconsensus sequence have a significantly higher efficiency asjudged by their capacity to inhibit cell fusion, it should beinteresting to test their therapeutic potential.

Materials and methods

Cell cultures and virusesBHK21-F, CV-1, MDBK and primary chick embryo cells were grown inDulbecco's medium containing 10% fetal calf serum. The Rostock strainof fowl plague virus, influenza A/FPV/Rostock/34 (H7N 1) and the Dobsonvariant of the Dutch strain, influenza A/FPV/Dutch/27 (H7N7) were used.Seed stocks were grown in eggs. To obtain radioactive virus with uncleavedHA for the protease assays, the Dutch strain was grown in BHK2I-F cellsin the presence of [3-S]methionine and 0.1 I M ionophore A23187(Calbiochem, Frankfurt/M, Germany) added to reinforced Eagle's medium(REM) lacking calcium ions and depleted of methionine (Klenk et al.,1984;Garten et al., 1989). For inhibition assays the Rostock strain was propagatedin chicken embryo cells.

Expression of recombinant vaccinia viruses in CV-1 cellsConfluent CV-1 cell monolayers were infected either with VV-HAwt(C.Roberts, H.-D.Klenk and W.Garten, manuscript in preparation), or withVV-HAwt and VVhFUR (Bresnahan et al., 1990) or with VVhFUR, eachat a m.o.i. of four. At 18 h after infection, the cells were starved ofmethionine in MEM without methionine for 60 min, each cell culture waslabeled with 100 ItCi [35S]methionine for 60 min, and the radioactive labelwas chased by adding MEM containing 10 mM methionine for another60 min. Then, the media were removed and the cells were washed twicewith ice-cold phosphate buffered saline and lysed for immunoprecipitation.

ImmunoreactionsFor immunoblotting, samples were mixed with an equal volume of 2-foldconcentrated sample buffer (500 mM dithiothreitol, 4% SDS, 10% glycerol,62.5 mM Tris-HCI, pH 6.8, 0.2% bromophenol blue), boiled for 3 min,and subjected to electrophoresis on 12% polyacrylamide gels. The proteinswere transferred to nitrocellulose membranes by electrophoresis. The blotwas incubated with rabbit anti-furin serum at 1:2000 dilution. The boundantibodies were detected using biotinylated anti-rabbit IgG and streptavidin-biotinylated horse-radish peroxidase complex (Amersham-Buchler,Braunschweig, Germany) and 4-chloro-1-naphthol as substrate (Kurodaet al., 1986). For immunoprecipitation, 20 jAl samples were incubated witheither 2yI rabbit anti-FPV serum or 2 AI rabbit anti-furin serum (diluted1:10) and 25 al of a suspension of protein A-Sepharose CL-4B (Sigma,Deisenhofen, Germany) in 500 jAl RIPA buffer (1% Triton X-100, 0.1%SDS, 1% deoxycholate, 10 mM EDTA, 20 mM Tris-HCl, pH 7.6) at4°C for 16 h. Subsequently, the immune complexes were washed three timeswith RIPA buffer and suspended in sample buffer for SDS-PAGE.

Protease assaysFor fluorometric assays, synthetic peptide derivates mimicking basic cleavagesites have been used as substrates. Boc-RQRR-AMC (=N-ot-t-butyloxy-carbonyl-L-arginyl-L-glutaminyl-L-arginyl-L-arginine-7-amido4-methyl-coumarin) manufactured as customer synthesis and Z-RR-AMC (=N-ca-benzyloxycarbonyl-L-arginyl-L-arginine-7-amido-4-methylcoumarin) werepurchased from Bachem Biochemica, Heidelberg, Germany. Boc-RVRR-AMC (=N-a-t-butyloxycarbonyl-L-arginyl-L-valyl-L-arginyl-L-arginine-7-amido-4-methylcoumarin) was purchased from peptide Institute Inc., Tokyo,Japan. Each substrate was kept as 100 jM stock solution in 50 mM HEPESbuffer, pH 7.3, containing 10 mM CaCI2 and 1% n-octyglucoside.Fluorogenic substrates (200 u1 of stock solutions) were incubated withprotease samples in a total volume of 2010 Al for 2 h at 37'C. The liberatedamido-4-methylcoumarin was fluorometrically measured at an excitationat 380 nm and an emission at 460 nm (Julius et al., 1984). Protease activitieswere expressed by relative fluorescence based on 7-amido-4-methylcoumarinof the same concentration as the given substrate.When uncleaved HA was used as substrate, protease samples (5 Al) were

incubated with 5 1 [35S]methionine-labeled virus at 37°C for 30 min. Toprotect viral proteins from non-specific degradation, 0.5 A1 of lipid extract(see below) had to be added to incubation mixtures containing detergent.After SDS-PAGE under reducing conditions, the ratio of cleaved touncleaved HA was determined on fluorographies using a Bio-Rad Model620 Videodensitometer. Proteolytic activity was either measured as apercentage of final HA cleavage reached after 2 h incubation at 37°C, or

defined by HA cleaving units. One unit is defined as1 ng of cleaved HAobtained inI min at 37°C at a concentration of 0.2 Ag uncleaved HA per A1.

Extraction of lipid from MDBK cellsWet MDBK cells obtained from 10 dishes (14 cm in diameter) were stirredwith 20 ml chloroform and 10 ml methanol at0°C for 20 min andcentrifuged at 1500 g for 30 min. The extracted lipids were dried undernitrogen and suspended in 100 Al 50 M HEPES, pH 7.3, 10 mM CaCI,,2% n-octylglucoside.

Preparation of furin samples from MDBK cellsRoutinely, MDBK cells grown to confluency in 300 Petri dishes (14 cmdiameter) were scraped from the plastic, washed with PBS three times, andcollected by centrifugation at 500 g. Cells were suspended at0°C in RSDmedium containing 10 mM Tris-HCI, pH 7.4, and disrupted by 30 strokesin a Dounce homogenizer. Cell nuclei were removed by centrifugation at2000 g for 20mmn, and the microsomal fraction was pelleted in aSW28rotor (Beckman) at 100 000 g for 30 min. Afterwards the pellet wasresuspended in 4 ml 50 mM HEPES, pH 7.3, containing 10 mM CaC12and laid on a discontinuous sucrose gradient (3 ml 65%, 7 ml 60%, 6 ml45%, 6ml 40% and 7ml 25% w/w) in the same buffer. Subcellular fractionswere separated by ultracentrifugation in aSW28 rotor at 25 000 r.p.m. for16 h and identified by the following marker enzymes: 5' nucleotidase andalkaline phosphatase (Engstrom, 1961) for plasma membranes, arylsulfatase(Chang et al., 1981) and acid phosphatase (Engstrom, 1961) for lysosomes,galactosyltransferase (Fleischer et al., 1969) for Golgi membranes, andglucose-6-phosphatase (Aronson and Touster, 1974) for endoplasmicreticulum. The Golgi fraction containing the HA cleaving activity wassolubilized by adding I-o-n-octyl-j3-D-glucopyranoside (BoehringerMannheim, Germany) at a detergent to protein ratio of 1.7:1 and stirringfor 1 h at 4°C. After removal of insoluble material by centrifugation at1000 g, the supernatant containing HA cleaving activity was loaded on a5-25% (w/w) sucrose gradient containing 50 mM HEPES pH 7.3, 10 mMCaCI2 and 1% n-octylglucoside, and centrifuged in a SW41 rotor(Beckman) at 35 000 r.p.m. for 16 h. The gradient was fractionated, 50drops each fraction. Fractions containing HA cleaving activity were appliedto a DEAE-Sepharose CL-6B column (Pharmacia, 4 cm long, 1.5 cm indiameter), which was equilibrated with the HEPES-calcium-octylglucosidebuffer containing the protease inhibitors leupeptin (10 pM), trans-epoxysuccinyl-l-leucylamido (4-guanidino) butane (E64, Sigma) (10 pM),phenylmethylsulfonyl fluoride (PMSF) (0.4 mM) and pepstatin A (Sigma)(0.4 mM). The column was extensively washed with this buffer and theneluted with the same buffer containing 300 mM NaCI. The eluted proteasecontaining fractions, were unified and concentrated by using a Centricon30 microconcentrator (Amicon, Witten, Germany). The protease wasexcluded from the gel matrix. Two further steps of FPLC were used toisolate the HA cleaving enzyme from other proteolytic activities. First, HAcleaving fractions were applied to strong anionic exchange chromatographyon a Mono Q HR5/5 column, equilibrated with 50 mM HEPES pH 7.3,10 mM CaC12 and 1% n-octylglucoside. Active fractions, which wereeluted by a NaCI gradient (0-300 mM) containing the application buffer,were pooled, concentrated to a total volume of 200 pl by a Centricon 30microconcentrator, and passed through a second FPLC system consistingof a Sepharose 12 HR 10/30 column equilibrated with 50 mM HEPES,pH 7.3, 10 mM CaCI2, 300 mM NaCl containing 1% n-octylglucoside.Furin fractions were concentrated as described above and used for analysis.

Synthesis of peptidyl chloroalkylketonesAlanyl-lysyl-arginyl chloromethylketone (AKR-CMK), tyrosyl-alanyl-lysyl-arginyl chloromethylketone (YAKR-CMK), tyrosyl-alanyl-arginyl-alanyl-lysyl-arginyl chloromethylketone (YARAKR-CMK) were essentiallysynthesized as described by Kettner and Shaw (1981). The acylated derivatespalmitoyl-phenylalanyl-lysyl-arginyl chloroethylketone (palFAKR-CEK),decanoyl-arginyl-glutamyl-lysyl-arginyl chloromethylketone (dec-REKR-CMK), which contained 70% decanoyl-arginyl-glutamyl (-y-methylester)-lysyl-arginyl chloromethylketone, were prepared by related procedures(Wikstrom et al., 1989). All the syntheses of peptidyl chloromethylketoneswill be described elsewhere (H.Angliker, P.Wikstrom, E.Shaw, W.Brennanand R.S.Fuller, manuscript in preparation). The acylated peptidylchloroalkylketones were solubilized at 10 mM concentration indimethylsulfoxide (DMSO), the non-acylated ones in 1 mM HCI as stocksolutions and kept at -20°C until usage. lodination reactions of YARAKR-CMK with 1251 were commercially performed by Amersham-Buchler(Braunschweig, Germany) according to the method described by Rauberet al. (1988).

2413

A.Stieneke-Grober et al.

Inhibition experimentsFor in vitro inhibition tests, 1 sll inhibitor solution was added to 9 Al sampleand incubated for 60 min at 0°C, except for the peptidyl chloroalkylketones,which were incubated at 37°C for 30 min. To measure the residual proteaseactivity, 5 Al suspension of FPV with uncleaved HA was added and analyzedas described. For in vivo inhibition, inhibitors were dispersed by sonification(Branson sonifier, 50 watt, 10 s) mixed in a total volume of 500 /1Dulbecco's medium and added 1 h post-injection to cell cultures. Theinhibitors were added to each medium used in the experiments. For HAcleavage analyses, chicken embryo cells were labeled 5 h after infectionwith [ S]methionine (10 ,Ci/ml) for a period of 15 h. Viral proteins wereimmunoprecipitated from the cells and analyzed by SDS-PAGE. Fusionof BHK21-F cells induced by FPV was studied in the absence or presenceof peptidyl chloromethylketones at different concentrations added to mediumI h post-injection. After 6 h, the cells were treated for 2 min with RPMI1640 medium (Flow Laboratories, Irvine, Scotland) adjusted to pH 5.1 withHCI, which was then replaced by the original medium for 3 h at 37°C.After fixation with ethanol at 0°C for 5 min, cells were washed twice withwater and stained for 20 min with 1:10 diluted Giemsa solution (Merck,Darmstadt, Germany). The polykaryons were photographed under amicroscope at 200 x magnification.

AcknowledaementsWe thank S.Berghofer for excellent technical assistance and S.Fischbachfor expert secretarial help. The anti-furin antiserum was a gift of Dr P.Barr.This work was supported by the Deutsche Forschungsgemeinschaft (Ro202/7-1 and SFB 286), by the Bundesministerium fiir Forschung undTechnologie (FKZ 11-113-90), and NIH grant DK37274. Part of this workwas done in fulfilment of the requirements for the degree of PhD of A.S.G.

ReferencesAnderson,R.G.W. and Pathak,R.K. (1985) Cell, 40, 635-643.Aronson,N.N. and Touster,O. (1974) Methods Enzymol., 31, 90-102.Barr,P.J. (1991) Cell, 66, 1-3.Barrett,T., Clarke,D.K., Evans,S.A. and Rima,B.K. (1987) Virus Res.,

8, 373-386.Bosch,F.X., Garten,W., Klenk,H.-D. and Rott,R. (1981) Virology, 113,725-735.

Bresnahan,P.A., Leduc,R., Thomas,L., Thorner,J., Gibson,H.L.,Brake,A.J., Barr,P.J. and Thomas,G. (1990) J. Cell Biol., 111,

2851 -2859.Chang,P.L., Rosa,N.E. and Davidson,R.G. (1981) Anal. Biochem., 117,

382-389.Collins,P.C., Huang,A.T. and Wertz,G.W. (1984) Proc. Natl. Acad. Sci.

USA, 81, 7683-7687.De Curtis,I. and Simons,K. (1989) Cell, 58, 719-727.Elango,N., Satake,M., Coligan,J.E., Norrby,E., Camargo,E. and

Venkatesan,S. (1985) Nucleic Acid Res., 13, 1559-1574.Engstrom,L. (1961) Biochim. Biophys. Acta, 52, 36-48.Fleischer,B., Fleischer,S. and Ozawa,H. (1969) J. Cell Biol., 43, 59-79.Fuller,R.S., Sterne,R.E. and Thorner,J. (1988) Annu. Rev. Physiol., 50,345-362.

Fuller,R.S., Brake,A.J. and Thorner,J. (1989) Science, 246, 482-486.Garten,W., Stieneke,A., Shaw,E., Wikstrom,P. and Klenk,H.-D. (1989)

Virology, 172, 25-31.Griffiths,G. and Simons,K. (1986) Science, 234, 438-443.Gotoh,B., Ogasawara,T., Toyoda,T., Inocencio,N.M., Hamaguchi,M. and

Nagai,Y. (1990) EMBO J., 9, 4189-4195.Hatsuzawa,K., Hosake,M., Nakagawa,T., Nagase,M., Shoda,A.,

Murakami,K. and Nakayama,K. (1990) J. Biol. Chem., 265,22075-22078.

Hosaka,M., Nagahama,M., Kim,W.-S., Watanabe,T., Hatsuzawa,K.,Ikemizu,J., Murakami,K. and Nakayama,K. (1991) J. Biol. Chem., 266,12127-12130.

Julius,D., Brake,A., Blair,L., Kunisawa,R. and Thorner,J. (1984) Cell,37, 1075-1089.

Kawaoka,Y. and Webster,R.G. (1988) Proc. Natl. Acad. Sci. USA, 85,321 -328.

Khatchikian,D., Orlich,M. and Rott,R. (1989) Nature, 340, 156-157.Keller,P.M., Davidson,A.J., Lowe,R.S., Bennett,C.D. and Ellis,R.W.

(1986) Virology, 152, 181-191.Kettner,C. and Shaw,E. (1981) Methods Enzymol., 80, 826-842.Klenk,H.-D. and Rott,R. (1988) Adv. Virus Res., 34, 247-281.Klenk,H.-D., Wollert,W., Rott,R. and Scholtissek,C. (1974) Virology, 57,

28-41.

2414

Klenk,H.-D., Garten,W. and Rott,R. (1984) EMBO J., 3, 2911-2915.Kuroda,K., Hauser,C., Rott,R., Klenk,H.-D. and Doerfler,W. (1986)EMBO J., 5, 1359-1365.

Lazarowitz,S.G. and Choppin,P.W. (1975) Virology, 68, 440-454.Li,S., Orlich,M. and Rott,R. (1990) J. Virol., 64, 3297-3303.Morrison,T., Ward,L.J. and Semerjian,A. (1985) J. Virol., 53, 851-857.Ogasawara,T., Gotoh,B., Suzuki,H., Asaka,I., Shimokata,K., Rott,R. and

Nagai,Y. (1992) EMBO J., 11, 467 -472.Ohuchi,M., Orlich,M., Ohuchi,R., Simpson,B.E.J., Garten.W., Klenk,H.-D.

and Rott,R. (1989) Virology, 168, 274-280.Ohuchi,R., Ohuchi,M., Garten,W. and Klenk,H.-D. (1991) J. Virol., 65,

3530-3537.Perez,L. and Hunter,E. (1987) J. Virol., 61, 1609-1614.Randolph,V.B., Winkler,G. and Stollar,V. (1990) Virology, 174,450-458.Ratner,C. et al. (1985) Nature, 313, 277-284.Rauber,P., Wikstrom,P. and Shaw,E. (1988) Anal. Biochem., 100,259-264.

Rice,C.M., Lenches,E.M., Eddy,S.L., Shin,S.J., Sheets,R.L. andStrauss,J.H. (1985) Science, 229, 726-733.

Richardson,C., Hull,D. Greer,P., Hasel,K., Berkovich,A., Englund,G.,Bellini,W., Rima,B. and Lazzerini,R. (1986) Virology, 155, 508-523.

Rott,R., Orlich,M., Klenk,H.-D., Wang,M.L., Skehel,J.J. and Wiley,D.C.(1984) EMBO J., 3, 3329-3332.

Sakaguchi,T., Matsuda,Y., Kiyokage,R., Kawahara,N., Kiyotani,K.,Katunuma,N., Nagai,Y. and Yoshida,T. (1991) Virology, 184, 504-512.

Seidah,N.G., Gaspar,L., Miou,P., Marcinkiewicz,M., Mbikay,M. andChretien,M. (1990) DNA Cell Biol., 9, 415-424.

Seidah,N.G., Marcinkiewicz,M., Benjannet,S., Gaspar,L., Beaubien,G.,Mattei,M.G., Lazure,C., Mbikay,M. and Chretien,M. (1991) Mol.Endocrinol., 5, 111-122.

Shaw,E. (1988) J. Biol. Chem., 263, 2768-2772.Shennan,K.I.J., Smeekens,S.P., Steiner,D.F. and Docherty,K. (1991) FEBS

Lett., 284, 277-280.Smeekens,S.P. and Steiner,D.F. (1990) J. Biol. Chem., 265, 2997 -3000.Smeekens,S.P., Avruch,A.S., LaMendola,J., Chan,S.J. and Steiner,D.F.

(1991) Proc. Natl. Acad. Sci. USA, 88, 340-344.Spaete,R.R., Saxena,A., Scott,P.I., Song,G.J., Probert,W.S., Britt,W.J.,

Gibson,W., Rasmussen.L. and Pachl,C. (1990)J. Virol., 64, 2922-2931.Sugrue,R.J., Bahadur,G., Zambou,M.C., Hall-Smith,M., Douglas,A.R.

and Hay,A.J. (1990) EMBO J., 9, 3469-3476.Tashiro,M., Ciborowski,P., Klenk,H.-D., Pulverer,G. and Rott,R. (1987)

Nature, 325, 536-537.Vey,M., Orlich,M., Adler,S., Klenk,H.-D., Rott,R. and Garten,W. (1992)

Virology, in press.Waxham,M.N., Server,A.C., Goodman,H.M. and Wolinsky,J.S. (1987)

Virology, 159, 381-388.Wikstrom,P., Kirschke,H., Stone,S. and Shaw,E. (1989) Arch. Biochem.,

Biophys., 270, 286-293.Wise,R.J., Barr,P.J., Wong,P.A., Kiefer,M.C., Brake,A.J. and

Kaufman,R.J. (1990) Proc. Natl. Acad. Sci. USA, 87, 9378-9382.Yamada,A., Tackeuchi,K. and Hishiyama (1988) Virology, 165, 268- 273.

Received on March 6, 1992